Computational and Experimental Insights into Asymmetric Rh-Catalyzed Hydrocarboxylation with CO2

Article abstract of DOI:10.1002/ejoc.202001469

The asymmetric Rh-catalyzed hydrocarboxylation of α,β-unsaturated carbonyl compounds was originally developed by Mikami and co-workers but gives only moderate enantiomeric excesses. In order to understand the factors controlling the enantioselectivity and to propose novel ligands for this reaction, we have used computational and experimental methods to study the Rh-catalyzed hydrocarboxylation with different bidentate ligands. The analysis of the C?CO₂ bond formation transition states with DFT methods shows a preference for outer-sphere CO₂ insertion, where CO₂ can undergo a backside or frontside reaction with the nucleophile. The two ligands that prefer a frontside reaction, StackPhos and ^tBu-BOX, display an intriguing stacking interaction between CO₂ and an N-heterocyclic ring of the ligand (imidazole or oxazoline). Our experimental results support the computationally predicted low enantiomeric excesses and highlight the difficulty in developing a highly selective version of this reaction.

Full text of DOI:10.1002/ejoc.202001469

Full Papers
doi.org/10.1002/ejoc.202001469
1
2
3
4
5
6
7
8
9
Computational and Experimental Insights into Asymmetric
Rh-Catalyzed Hydrocarboxylation with CO2
Ljiljana Pavlovic,[a] Martin Pettersen,[b] Ashot Gevorgyan,[b] Janakiram Vaitla,[b]
Annette Bayer,*[b] and Kathrin H. Hopmann*[a]
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
The asymmetric Rh-catalyzed hydrocarboxylation of α,β-unsatu-
rated carbonyl compounds was originally developed by Mikami
and co-workers but gives only moderate enantiomeric excesses.
In order to understand the factors controlling the enantiose-
lectivity and to propose novel ligands for this reaction, we have
used computational and experimental methods to study the
Rh-catalyzed hydrocarboxylation with different bidentate li-
gands. The analysis of the CÀ CO2 bond formation transition
states with DFT methods shows a preference for outer-sphere
CO2 insertion, where CO2 can undergo a backside or frontside
reaction with the nucleophile. The two ligands that prefer a
frontside reaction, StackPhos and Bu-BOX, display an intriguing
stacking interaction between CO2 and an N-heterocyclic ring of
the ligand (imidazole or oxazoline). Our experimental results
support the computationally predicted low enantiomeric ex-
cesses and highlight the difficulty in developing a highly
selective version of this reaction.
t
Introduction
Interestingly, many of the known CÀ CO2 bond formations
result in generation of chiral carboxylic acids, but as racemic
mixtures only.[2b,e,h] Indeed, the design of enantioselective
CÀ CO2 bond formation reactions remains a major challenge.
This is demonstrated by the fact that only very few studies on
asymmetric CÀ CO2 bond formation have been reported.[1f,2c,3] In
order to broaden the usefulness of CO2 as a carbon synthon in
the chemical and pharmaceutical industry, it is essential that
novel enantioselective carboxylation protocols are developed,
for example for the preparation of chiral carboxylic acids, which
are important intermediates in many synthetic processes.[4]
A promising asymmetric CÀ CO2 bond formation protocol
has been reported by Mikami and co-workers in 2016, involving
the first enantioselective hydrocarboxylation of α,β-unsaturated
carbonyl compounds (Figure 1).[2c] The rhodium-based reaction
involved the use of (S)-SEGPHOS as a chiral ligand, but only
moderate enantiomeric excesses (e.e.’s) of up to 66% could be
Widespread efforts are currently devoted to the search of
catalysts, which can fixate CO2 into organic molecules.[1]
A
significant part of this activity is focused on metal-catalyzed
carbon-carbon bond formation with CO2.[2] For the metal-
catalyzed formation of saturated carboxylic acids, different
protocols have been reported, including carboxylation of
halides (CÀ X bonds)[2a,b] and reductive carboxylation of unsatu-
rated compounds such as alkenes.[2c–h] An example of the
carboxylation of Csp3À X bonds has been reported by Martin
and co-workers, who developed a mild Ni(I)-catalyzed protocol
for converting benzyl halides and CO2 to phenylacetic acids.[2b]
The catalytic reductive carboxylation of alkenes is a challenging
area, which has witnessed some progress in recent years. For
example, Greenhalgh and Thomas reported a Fe(II)-catalyzed
synthesis of α-aryl carboxylic acids from styrene derivatives and
CO2.[2e] A Cu(I)/CsF-based protocol for the incorporation of CO2
into disubstituted alkenes was reported by Skrydstrup, Nielsen,
and co-workers.[2h]
[a] Dr. L. Pavlovic, Prof. Dr. K. H. Hopmann
Hylleraas Center for Quantum Molecular Sciences
Department of Chemistry, UiT The Arctic University of Norway
9037 Tromsø, Norway
E-mail: kathrin.hopmann@uit.no
https://site.uit.no/choco
[b] M. Pettersen, Dr. A. Gevorgyan, Dr. J. Vaitla, Prof. Dr. A. Bayer
Department of Chemistry, UiT The Arctic University of Norway
9037 Tromsø, Norway
E-mail: annette.bayer@uit.no
https://site.uit.no/bayerlab/
Supporting information for this article is available on the WWW under
https://doi.org/10.1002/ejoc.202001469
© 2020 The Authors. European Journal of Organic Chemistry published by
Wiley-VCH GmbH. This is an open access article under the terms of the
Creative Commons Attribution License, which permits use, distribution and
reproduction in any medium, provided the original work is properly cited.
Figure 1. Enantioselective hydrocarboxylation reaction reported by Mikami
and coworkers.[2c]
Eur. J. Org. Chem. 2021, 663–670
663
© 2020 The Authors. European Journal of Organic Chemistry
published by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202001469
achieved.[2c] The (S)-BINAP ligand gave similar results to (S)- Results and Discussion
SEGPHOS whereas other ligands, such as (S)-SynPhos or
1
2
3
(R,R)-iPr-DuPhos, provided significantly lower e.e.’s.[2c]
Our study of the Rh-catalyzed asymmetric hydrocarboxylation
A computational analysis of the related non-enantioselec-
tive Rh-COD-catalyzed hydrocarboxylation reaction showed that
during CÀ CO2 bond formation, the CO2 molecule does not
interact with rhodium.[5] Moreover, it was shown that benzylic
substrates display an unusual η6-coordination mode, with the
nucleophilic carbon positioned up to 3.6 Å away from
rhodium.[5] The same substrate binding mode and preference
for an outer sphere CO2 insertion were found computationally
for the chiral Rh-(S)-SEGPHOS catalyst.[6] This raises the question
how the enantioselectivity is controlled in systems where CO2 is
not constrained through interactions with the metal. Although
CO2 preferably is positioned in the outer sphere, it may still be
affected by repulsive and attractive nonbonding interactions
with the ligand. A better understanding of the factors that
govern the preferred positions and orientations of CO2 may
help to design catalysts with higher enantioselectivities.
Modern computational methods are sufficiently advanced
to provide insights into the factors that control the enantiose-
lectivity in metal-catalyzed reactions.[7] For example, the
selectivity may be influenced by the presence of specific
interactions between the chiral catalyst and the substrate, and
in particular, nonbonding forces may contribute significantly to
the preferred formation of one product enantiomer.[7–8] The
identification of the selectivity-determining interactions typi-
cally relies on the computational optimization of the involved
diastereomeric transition states. Such structures are generally
built manually, followed by DFT optimizations, using different
optimization algorithms.[9] However, approaches to speed-up
the computational analysis through automatized techniques
have been put forward,[10] with one example being the open-
source toolkit AARON (An Automated Reaction Optimizer for
New catalysts) designed by Wheeler and co-workers.[10a] AARON
employs TS templates provided by the user, but can automati-
cally swap the ligands to build new geometries.
reaction consists of three parts. Initially, we validated the
computational protocol through analysis of the Rh-(S)-SEG-
PHOS-catalyzed hydrocarboxylation of two experimentally
known substrates.[2c] Next, we expanded our computational
study to include the CO2 insertion TSs for three additional chiral
ligands, which have not been used in experiments on this
reaction. Finally, we conducted an experimental evaluation of
the corresponding Rh-complexes for hydrocarboxylation of
ethyl 2-phenylacrylate.
For the analysis of the chiral ligands, 10 outer sphere CO2
insertion TSs were built for each ligand, with different ligand-
substrate orientations (Figure 2). Five of them were pro-(S)-TSs,
and five the corresponding pro-(R) TSs. In the conformations
TS1a and TS1b, the phenyl ring of the substrate interacts with
the Rh-center in an η6 fashion, whereas CO2 is in the outer
sphere, leading to a backside CÀ CO2 bond formation (reminis-
cent of a SE2(back) reaction). The difference between TS1a and
TS1b is the orientation of the ester moiety (Figure 2). At TS2a
and TS2b, the substrate is still bound in an η6 fashion, but the
CO2 is positioned closer to metal, leading to a frontside reaction
(reminiscent of a SE2(front) reaction). At TS3, both the phenyl
group and the carbonyl oxygen of the substrate interact with
the Rh-center. It is important to highlight that for the
comparative analysis of the four ligands, only outer sphere CO2
insertion was considered,[5] because the TS conformations,
where interactions between Rh and CO2 take place (referred to
as inner sphere CO2 insertion), show very high barriers (TS4_S
and TS4_R, Supporting Information, Table S1). The four studied
chiral ligands are given in Figure 3.
Computational analysis of Rh-(S)-SEGPHOS: The Rh-SEG-
PHOS-catalyzed hydrocarboxylation was here investigated com-
putationally with the styrene-type α,β-unsaturated carbonyl
substrates sub1 and sub2 (Figure 1), which previously have
been studied experimentally by Mikami and co-workers.[2c] The
overall hydrocarboxylation mechanism for substrates of this
type has been reported with [Rh(cod)Cl]2 (SI, Figure S1).[5] We
have here studied the full mechanism with Rh-(S)-SEGPHOS as
the catalyst and methyl 2-phenylacrylate (sub1) as the sub-
strate, with the energy profile shown in Figure S2 (Supporting
Information). The mechanistic steps include a transmetallation
of an ethyl from diethylzinc to the precatalyst, followed by a β-
hydride elimination to give an Rh-H-Et intermediate. Insertion
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Herein, we perform
a computational analysis of the
selectivity-determining factors in the Rh-catalyzed hydrocarbox-
ylation for four chiral rhodium complexes, of which three
ligands have not previously been tested in this reaction. Ligand
swapping is performed with AARON, followed by DFT optimiza-
tions. To validate the enantioselectivities predicted by the
computations, an experimental analysis of all systems is
performed.
Figure 2. Five TS orientations considered here. For each of these, both pro-(R) and pro-(S) conformations were included.
Eur. J. Org. Chem. 2021, 663–670
www.eurjoc.org
664
© 2020 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202001469
Carboxylation of methyl 2-phenylacrylate: In order to validate
1
2
3
4
5
6
7
8
9
our computational protocol and our mechanistic understanding
of this reaction, we first analyzed the Rh-(S)-SEGPHOS-catalyzed
CÀ CO2 bond formation with sub1 (Figure 1). The results support
our previous observation that CO2 prefers to be in the outer
sphere during CÀ CO2 bond formation,[5] as the inner and outer
sphere TSs with Rh-(S)-SEGPHOS show an energy difference of
17.3 kcal/mol in favor of outer-sphere insertion (SI, Table S1,
Figure S3).
At the lowest-lying outer sphere transition state TS1a_Ssub1/L1
,
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
the η6-coordinated enolate attacks CO2 via its re face ( G =
12.1 kcal/mol relative to the Rh-benzyl intermediate, Figure S4,
SI) and the experimentally observed (S)-product is obtained. At
TS1a_Rsub1/L1, which is higher in energy by 0.7 kcal/mol, CO2 is
attacked by the enolate si face, giving the (R)-product (Figure 5).
Other outer sphere conformations (Figure 2) were significantly
higher in energy (Table 1). On the basis of all computed TS
energies, we evaluated the e.e. for the Rh-(S)-SEGPHOS-catalyzed
hydrocarboxylation of sub1, providing a computed e.e. of 53.8%
(S), in very good agreement with the experimentally reported e.e.
of 60.0% (S).[2c]
¼
~
Figure 3. Four chiral ligands studied here in Rh-catalyzed hydrocarboxyla-
tion.
of the substrate leads to an energetically low-lying Rh-benzyl
species that can attack CO2.[5] The CO2 insertion is rate- and
enantioselectivity-determining.[5] At the carboxylation TS, the
benzyl group prefers to coordinate in an η6 mode to rhodium,
with the formally negative charge on the substrate delocalized
between the nucleophilic carbon and the ester group, yielding
an intermediate enolate (Figure 4). The enolate can attack CO2
from its re or si face, and with a chiral ligand, unequal amounts
of the (R)- and (S)-enantiomer of the product can be formed.
Various noncovalent interactions between the ligand and
sub1 can be identified at the two energetically lowest-lying
SEGPHOS TSs, TS1a_Ssub1/L1, and TS1a_Rsub1/L1 (Figure 5). At
TS1a_Ssub1/L1, the phenyl rings of SEGPHOS form two CÀ H···π
interactions (2.95, 3.10 Å) with the phenyl of the substrate. At
the energetically higher lying TS1a_Rsub1/L1, SEGPHOS forms
three CÀ H···π interactions with sub1, two with the substrate
phenyl (2.97 and 3.14 Å), and one with the methyl group of the
ester moiety (3.16 Å, Figure 5). As the strength of these CÀ H···π
interactions appear similar at the two diastereomeric TSs, they
do not seem to determine the selectivity. An analysis of CÀ H···O
attractions at the two TSs shows comparable distances for
interactions within the substrate (TS1a_Ssub1/L1: 2.16 Å, TS1a_
R
sub1/L1: 2.11 Å), but significant differences in the intermolecular
CÀ H···O interaction between the sub1 carbonyl and the
SEGPHOS phenyl (TS1a_Ssub1/L1: 2.46 Å, TS1a_Rsub1/L1: 3.00 Å). We
speculate that this CÀ H···O interaction may be an essential
factor in determining the enantioselectivity in the Rh-(S)-
SEGPHOS-catalyzed hydrocarboxylation of methyl 2-phenylacry-
late.
If CO2 is placed closer to rhodium, here referred to as
frontside insertion (TS2, Figure 2), the barriers increase by
several kcal/mol (Figure 5). Interestingly, the frontside attack
provides an incorrect enantioselectivity, as the TS2a Rsub1/L1
Figure 4. Illustration of the enolate intermediate of sub1 and its attack on
CO2.
¼
Table 1. Barrier differences (ΔΔG , kcal/mol, 273 K) for different TS conformations (Figure 2) in Rh-catalyzed hydrocarboxylation of sub1.
η6, backside
TS1a_S
η6, frontside
TS2a_S
η2, backside
e.e.comp
[%]
e.e.exp
[%]
Ligand
TS1a_R
TS1b_S
TS1b_R
TS2a_R
TS2b_S
TS2b_R
TS3_S
TS3_R
L1 (SEGPHOS)
L2 (StackPhos)
0.0
2.2
0.7
2.8
3.1
2.1
2.0
3.0
6.5
4.0
7.3
0.8
4.9
1.0
8.3
15.2
7.9
10.8
53.8 (S)
47.0 (S)
60.0 (S)[c]
0.0[a],
0.8[b]
0.0
0.6,[a]
1.9[b]
0.5
n.d.[d]
L3 (tBu-BOX)
L4 (BDPP)
1.9
0.5
0.7
0.0
3.1
0.8
0.8
1.9
3.6
8.6
2.5
6.4
3.2
9.7
5.3
12.1
6.4 (S)
24.3 (R)
(0)[e]
(4)[e]
5.8
5.5
[a] TS2a structures as given in Figure 8 (TS2a_Ssub1/L2/TS2a_Rsub1/L2). [b] TS2a structures with stacking of pentafluorophenyl and phenyl as given in the SI,
Figure S6 (TS2a_stack Ssub1/L2/TS2a_stack_Rsub1/L2). [c] From ref.[2c]. [d] Only racemic StackPhos could be tested, and the e.e. could thus not be determined. [e]
Experimental results obtained here with sub3, which has an ethyl group instead of the methyl in sub1 (Figure 1).
Eur. J. Org. Chem. 2021, 663–670
www.eurjoc.org
665
© 2020 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202001469
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 5. Illustration of the noncovalent interactions at four of the optimized CO2 insertion TSs for Rh-(S)-SEGPHOS-catalyzed hydrocarboxylation of methyl 2-
phenylacrylate (sub1). Only some of the hydrogens are shown for clarity. Distances in Å.
structure is 2.5 kcal/mol lower in energy than TS2a Ssub1/L1. The
experimentally observed (S)-selectivity)[2c] is thus dominated by
the backside structures. These findings highlight the need to
compare computationally predicted TSs with experimental
selectivities to evaluate if appropriate TS conformations were
located.
The TS3 conformations, where the ester of the substrate
interacts with rhodium (Figure 2), are ~8 kcal/mol higher in
energy than TS1 and are not considered relevant (Table 1).
Carboxylation of 4-(tert-butyl)benzyl 2-phenylacrylate: We
proceeded to analyze sub2, which contains two phenyl rings
(Figure 1), leading to several favorable CÀ H…π interactions
Figure 6. Illustration of the preferred TSs for Rh-(S)-SEGPHOS-catalyzed
carboxylation of sub2. Only some of the hydrogens are shown for clarity.
Distances in Å.
during CÀ CO2 bond formation (Figure 6). A similar pattern as for
sub1 is observed, where at the lowest-lying transition state
¼
~
TS1a_Ssub2/L1 ( G =12.0 kcal/mol), the Rh-benzyl (SI, Figure S4)
attacks CO2 from its re face, resulting in the (S)-product. A
favorable CÀ H···O (2.47 Å) interaction is seen at TS1a_Ssub2/L1 but
lacks at TS1a_Rsub2/L1, which is higher in energy by 1.0 kcal/mol.
The computed e.e. of 73% (S) is in good agreement with the
experimental value of 66% (S).[2c]
the chiral catalyst is promoting the enantioselectivity through
the positioning of the alkene substrate, not through interac-
tions with CO2.
The combined results for sub1 and sub2 indicate that the
enantioselectivity of Rh-(S)-SEGPHOS-catalyzed hydrocarboxyla-
tion appears to be a result of favorable CÀ H···O interactions
between the substrate and the SEGPHOS ligand. At the
preferred TS1a conformations (Figure 5 and Figure 6), the CO2
molecule is placed away from the metal center (>5 Å) and thus
Potential of other ligands in the Rh-catalyzed asymmetric
hydrocarboxylation: We selected a set of ligands structurally
different from SEGPHOS from the library of AARON[10a] (L2–L4,
Figure 3) and investigated their predicted enantioselectivities
with DFT. The set includes one P,N ligand (L2: StackPhos),[11] an
t
N,N ligand (L3: Bu-BOX)[12] and a P,P ligand (L4: BDPP).[13] These
Eur. J. Org. Chem. 2021, 663–670
www.eurjoc.org
666
© 2020 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202001469
ligands have shown good performance in other asymmetric
therefore discussed in further detail here. Besides the CO2-
imidazole stacking, the lowest lying TS2a_Ssub1/L2 also displays
an intriguing F-π attraction between a fluoro group of the
pentafluoro-phenyl and the naphthalene ring (3.05 Å), along-
side a CÀ H···F interaction (2.53 Å, Figure 8). Similar F-π inter-
actions to phenantrene-like aromatic systems have been
reported in the literature.[17] Interestingly, this F-π interaction is
not seen in the X ray structure of the StackPhos ligand,[11a]
which instead displays π-π stacking between pentafluorophenyl
and naphthalene subunits (3.38 Å). In our computations, this π-
π stacking increases the TS energy by 2.5 kcal/mol (SI, Fig-
ure S5).
An alternative π-π interaction between pentafluorophenyl
and another phenyl substituent increases the CO2 insertion
barrier slightly by 0.8 kcal/mol (TS2a_stack_Ssub1/L2 SI, Figure S6).
In the case of backside insertion with StackPhos, the imidazole-
CO2 interactions are absent, which increases the barriers by 2 to
3 kcal/mol (Table 1). The TS3 structures, where the ester
carbonyl interacts with rhodium, are more than 11 kcal/ mol
above the TS2 structures and therefore are not relevant.
The best (R)-pathway obtained for sub1 with StackPhos
proceeds via frontside insertion and is 0.6 kcal/mol above the
best (S)-structure (TS2a_RsubI/L2, Figure 8). This TS also displays
stacking of CO2 above the imidazole moiety and an F-π
interaction between pentafluorophenyl and the naphthalene
subunits (Figure 8). The e.e. computed on the basis of all
obtained StackPhos TS structures is 47% (S) (Table 1), which
indicates that this ligand is not expected to perform signifi-
cantly better than SEGPHOS.
1
2
3
4
5
6
7
8
9
transformations (allylation, aziridination, hydrovinylation),[14] and
to our knowledge, they have not previously been used for Rh-
catalyzed hydrocarboxylation.
The outer sphere TS conformations depicted in Figure 2
were evaluated for L2–L4 and sub1 through manual DFT
calculations, with the energies summarized in Table 1 (geo-
metric parameters are shown in Figure 7, Figure 8 and
Tables S1–4, SI). For BDPP (L4), we see a similar behaviour as for
SEGPHOS, with a preference for backside insertion (Table 1).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
t
However, the StackPhos (L2) and Bu-BOX (L3) ligands show a
computed preference for frontside insertion. Both ligands
display an intriguing stacking interaction between CO2 and the
N-heterocyclic ring of the ligand (imidazole or oxazoline,
Figure 7, SI, Figure S7).
It can be noted that related attractive stacking interactions
have been predicted in computational studies focusing on the
binding of CO2 to N-heterocyclic compounds,[15] and in exper-
imental and computational studies on the solvation of aromatic
compounds in supercritical CO2.[16] However, to our knowledge,
the heterocycle-CO2 stacking interaction has not been described
in the context of an organometallic ligand or a CO2 insertion
reaction.
The heterocycle-CO2 interaction appears strongest at the
StackPhos TS geometries, with a nitrogen-CCO2 distance of
3.22 Å (Figure 7). The StackPhos TS geometries with sub1 are
The other studied ligands are predicted to give low e.e.’s.
Our calculations show that with the (R,R)-tBu-BOX chiral ligand,
at the lowest-lying TS2a S sub1/L3, the frontside CO2 insertion is
preferred (SI, Figure S7). The opposite enantiomer TS2a R
sub1/L3
is higher in energy by only 0.5 kcal/mol. The predicted e.e. on
the basis of all optimized TS conformations is only 6.4%
(Table 1).
With the (R,R)-BDPP ligand, at the lowest-lying TS1a R
the CO2 prefers backside insertion (SI, Figure S7). TS1a_S
has a barrier that is only 0.5 kcal/mol higher than TS1a_R
,
sub1/L4
sub1/L4
Figure 7. Stacking of CO2 above the N-heterocyclic ring of L2 and L3 at the
frontside TSs. Distances in Å.
.
sub1/L4
Figure 8. Illustration of the preferred TSs for Rh-(R)-StackPhos-catalyzed carboxylation of sub1. Distances in Å
Eur. J. Org. Chem. 2021, 663–670
www.eurjoc.org
667
© 2020 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202001469
The TSs for the frontside CO2 insertion are higher in energy by
the asymmetric rhodium-catalyzed hydrocarboxylation of acryl-
ates.
Our DFT analysis of the mechanism supports a preference
1
2
3
4
5
6
7
8
9
more than 5 kcal/mol (Table S4). This scenario is reminiscent of
the biphospine ligand (S)-SEGPHOS. These observations may be a
consequence of the bulky phenyl groups of the ligands, which
restrict CO2, making the backside insertion more preferable. The
computed e.e. for this ligand is 24% (R) (Table 1).
Experimental analysis of Rh-catalyzed hydrcarboxylation
of L1 to L4: We analyzed the ability of L1 to L4 to mediate the
CO2 insertion reaction with sub3 (Figure 1), which is closely
related to the computationally studied substrate sub1, but
which has an ethyl instead of a methyl ester. In the work by
Mikami and co-workers, sub3 and sub1 behaved similarly,
providing respectively 66% and 60% e.e.’s for Rh-SEGPHOS
catalyzed hydrocarboxylation.[2c]
In our work, we obtained a product yield of 48% and an e.e.
of 32% with L1 and sub3 (Table 2). Although the yield is similar
as previously reported, the e.e. is somewhat lower than the
reported 66%.[2c] For L2, only a racemic mixture of the ligand
could be tested,[18] providing a yield of 74% for carboxylation of
sub3 (Table 2). Thus, L2 may provide reasonable yields, and
may be a relevant starting point for future development of
ligands for this reaction.
For L3, experimental hydrocarboxylation of sub3 gave the
acid in as much as 99% yield but with 0% e.e. (Table 2), in good
agreement with our predictions for sub1 of 6.4% e.e. (Table 1).
For L4, our experimental results on sub3 showed 94% yield,
but only 4% e.e. (Table 2), in line with the predicted low e.e. of
24% e.e. for sub1 (Table 1).
for an η6 coordination of benzylic substrates and an outer
sphere insertion of CO2 also with chiral ligands.[5] The reported
experimental enantioselectivity with SEGPHOS[2c] is reproduced
for substrates sub1 and sub2 in our calculations and is
predicted to arise from the CÀ H···O interaction between a
phenyl group of SEGPHOS and the carbonyl group of the
substrate.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Our computations on the chiral P,N ligand StackPhos (L2),
t
the N,N ligand Bu-BOX (L3) and the P,P ligand BDPP (L4)
t
showed up to 47% e.e. for sub1. For StackPhos and Bu-BOX,
the preferred transition state geometries display an intriguing
stacking interaction of CO2 with the N-heterocyclic ring
(imidazole or oxazoline, Figure 7). Experimental analyses of
ligands L1 to L4 showed that all are able to catalyze the
hydrocarboxylation reaction, with L2, L3, and L4 providing
good yields of 74 to 99% for carboxylation of sub3. Although
the experimentally observed enantiomeric excesses are low,
they are in good agreement with computations, underpinning
the ability of DFT-D to adequately model complex enantiose-
lective reactions.
Our combined results on Rh-catalyzed hydrocarboxylation
indicate that the enantioselectivity of this reaction is difficult to
control. A possible strategy to be considered is to steer CO2 into
a specific position to decrease its conformational freedom. The
noncovalent stacking interactions observed between CO2 and
L2 or L3 (Figure 7) may be interesting in this sense and variants
of these ligands may thus be a relevant starting point for future
developments.
We conclude that our experimental results are in good
agreement with the low e.e.’s predicted by the computations.
This validates the proposed outer sphere mechanisms pre-
sented here for ligands L1 to L4 and indicates that DFT-D
methods can be employed to model the enantioselectivities of
these kinds of systems. At the same time, it highlights the Computational section
difficulty to make a selective version of the rhodium-catalyzed
hydrocarboxylation of acrylates.
Computational models: Calculations were performed with full
substrates sub1 and sub2 (Figure 1) and with the full ligands
(Figure 3). No molecular truncations or symmetry constraints were
applied.
Conclusion
Computational methods: All calculations were performed at the DFT
level of theory as implemented in the Gaussian09 package.[19] For
geometry optimizations, the DFT functional PBE[20] was employed
We have employed computational and experimental methods
to study the potential of bidentate chiral ligands L1 to L4 for
together with the Grimme empirical dispersion correction (D2[21]
)
and the implicit polarizable continuum model using the integral
equation formalism, IEFPCM[22] (DMF solvent). The PBE functional
has been found to be an adequate choice for rhodium-catalyzed
hydrocarboxylation reactions in our previous study,[5] where it
provided a good agreement with experimental results.[2c] The
geometries of all intermediates and transition states were fully
optimized and frequency calculations were performed in order to
confirm the nature of the stationary points, where all transition
states structures exhibited only one imaginary frequency.
Table 2. Experimental yields and e.e’s with four chiral ligands employed in
Rh-catalyzed hydrocarboxylation of sub3.
In geometry optimizations, the BS1 basis set was employed,
consisting of 6-311G(d,p)[23] for C, H, O, N, F, and P, and the
LANL2DZ[24] basis set and pseudopotential for rhodium, including
an extra f polarization function with exponent 1.35.[25] A larger basis
set, BS2, was employed for single-point energy calculations,
consisting of 6-311+G(2d,2p) on all non-metal atoms and LANL2TZ
(f) on rhodium.
Ligand
Yields [%]
e.e.exp [%]
L1 (SEGPHOS)
L2 (StackPhos)
L3 (tBu-BOX)
L4 (BDPP)
48.0
74.0
99.0
94.0
32.0
n.d.[a]
0.0
4.0
[a] n.d=Not detected.
Eur. J. Org. Chem. 2021, 663–670
www.eurjoc.org
668
© 2020 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202001469
°
In order to convert computed free energies (ΔG , BS1) at 1 atm into
1
2
3
a 1 M standard state, a standard state (SS) correction was included.
At 273 K, this correction is À 1.69 kcal/mol (for a reaction that goes
from 2 moles to 1).[26]
4
5
6
7
The final Gibbs free energy was determined with the following
°
°
expression: ΔG 1M,273K =ΔG 1atm,BS1,273KÀ ΔEBS1 +ΔEBS2 +SS273K.
Figure 9. Experimental procedure for the preparation of Rh-complexes.
The enantiomeric excess (e.e.) was computed using the formula
Eq. (1):[7,27]
8
solution of [Rh(cod)Cl]2, which was accompanied by precipitation of
a white powder (AgCl/NaCl). The resulting mixture was stirred at
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
°
20 C for 1 h. Afterward, the precipitate was filtered off and the
where kRi are the computed rate constants of TS structures with (R)
configuration, which are summed from i=1 to i=n, where n is
equal to the number of TSs within 3 kcal/mol from the best TS. ksi is
the equivalent for (S)-TSs.
solvent was evaporated to give the corresponding complex as an
orange powder.
General experimental procedure for Rh-catalyzed hydrocarboxylation
of ethyl 2-phenylacrylate (Table 2): Inside of the glove box an oven-
dried 25 mL Schlenk flask was charged with corresponding Rh-
complex (10 mol%) and AgSbF6 (10 mol%). The flask was sealed
with a rubber septum, removed from the glove box, evacuated,
filled with CO2, and equipped with a CO2 balloon. This was followed
by sequential addition of dry DMF (5 mL) and ethyl 2-phenyl-
acrylate (150 mg, 1 equiv.) using syringes. The resulting mixture
was transferred into an ice bath where under vigorous stirring 1 M
solution of Et2Zn in hexane (1.2 equiv.) was added dropwise using a
AARON ligand swapping: The TS library used for AARON[10a] was
based on the SEGPHOS structures obtained in the manual DFT
analysis. Three ligands present in the AARON ligand library (L2, L3,
L4) were then specified to be swapped with SEGPHOS. We
preoptimized the conformations with the swapped ligands with
AARON in two steps, using HF/6-31 in the first step and PBE-D2/
BS1mod in the second step, where BS1mod is as BS1 but lacks the
additional f polarization function on rhodium, as AARON did not
allow the addition of basis functions. The obtained geometries for
all ligands were then used as input for further manual DFT
investigations, with the protocol as described above for manual
DFT calculations. Note that for L4, the (R,R) ligand was computed,
but the (S,S) ligand was used in experiments (which should give
opposite enantioselectivity).
°
syringe. The resulting mixture was allowed to stir at 0 C for 3 h.
Then the reaction mixture was diluted with Et2O (5 mL) and
carefully neutralized using 6 M HCl (5 mL). The acidic solution was
diluted with water (5 mL) and removed using a separating funnel.
The organic phase was then extracted using a solution of saturated
NaHCO3 (3×30 mL). The collected aqueous solution was carefully
treated with 6 M HCl (60 mL) and extracted using Et2O (3×30 mL).
Collected Et2O solution was washed with distilled water (30 mL)
and evaporated to give the target acid as a faint orange oil.
Enantiomers were separated using SFC on a chiral column (CEL-2),
eluent iPrOH:EtOH:TFA – 70:30:2, and gradient 3–8, 10 min run.
Experimental Section
Experimental Details: Commercially available starting materials,
reagents, catalysts, and anhydrous and degassed solvents were
used without further purification. Thin-layer chromatography was
carried out using Merck TLC Silica gel 60 F254 and visualized by
short-wavelength ultraviolet light or by treatment with potassium
Starting from 0.851 mmol of ethyl 2-phenylacrylate the product
was obtained as a faint orange oil, yield 48%, e.e. 32% (0.091 g, [Rh
(cod)(((S)-SEGPHOS)]SbF6), yield 74% (0.121 g, [Rh(cod)((rac)-Stack-
Phos)]SbF6), yield 99%, e.e. 0% (0.189 g, [Rh(cod)((S,S)-tBu-BOX)]
SbF6), yield 94%, e.e. 4% (0.178 g, [Rh(cod)((S,S)-BDPP)]SbF6). 1H
NMR (400 MHz, CDCl3): δ=10.38 (br s, 1H), 7.39–7.24 (m, 5H), 4.21
(q, J=7.1 Hz, 2H), 1.87 (s, 3H), 1.22 (t, J=7.1 Hz, 4H). 13C NMR
(101 MHz, CDCl3): δ=177.0, 171.9, 137.7, 128.4, 128.0, 127.4, 62.3,
58.7, 22.0, 14.0.
1
permanganate (KMnO4) stain. H, 13C, 19F, and 31P NMR spectra were
1
°
recorded on a Bruker Avance 400 MHz at 20 C. All H NMR spectra
are reported in parts per million (ppm) downfield of TMS and were
measured relative to the signals for CHCl3 (7.26 ppm). All 13C NMR
spectra were reported in ppm relative to residual CDCl3 (77.20 ppm)
1
and were obtained with H decoupling. Coupling constants, J, are
reported in Hertz (Hz). High-resolution mass spectra (HRMS) were
recorded from methanol solutions on an LTQ Orbitrap XL (Thermo
Scientific) in positive electrospray ionization (ESI) mode.
Acknowledgements
(S)-SEGPHOS, (S,S)-tBu-BOX, and (S,S)-BDPP ligands are commercially
available. Ethyl 2-phenylacrylate, StackPhos, and corresponding Rh
complexes were prepared according to slightly modified literature
procedures. For more details, see Electronic Supporting Informa-
tion.
This work has been supported by the Research Council of Norway
(No. 262695, No. 300769), by the Tromsø Research Foundation
(No. TFS2016KHH), by Notur – The Norwegian Metacenter for
Computational Science through grants of computer time (No.
nn9330k and nn4654k), and by NordForsk (No. 85378). We thank
Manuel K. Langer for support with the SFC and Prof. Steven
Wheeler, Victoria M. Ingman, Anthony J. Schaefer, and Stig Rune
Jensen for advice and technical assistance in the implementation
of AARON
General experimental procedure for the preparation of Rh-complexes
(Figure 9): Inside of the glove box an oven-dried 25 mL round
bottom flask was charged with [Rh(cod)Cl]2 (100.0 mg, 1 equiv.) and
AgSbF6.The flask was sealed with a rubber septa, removed from the
glove box, and equipped with an Ar balloon. Inside of the glove
box, another oven dried 25 mL round bottom flask was charged
with the corresponding chelating ligand (2 equiv.), sealed with a
rubber septum, removed from the glove box, and equipped with
an Ar balloon. Both flasks were charged with dry CHCl3 (5 mL) and
°
allowed to stir for 30 min at 20 C. This was followed by the
dropwise addition of CHCl3 solution of the ligand to the stirring
Eur. J. Org. Chem. 2021, 663–670
www.eurjoc.org
669
© 2020 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH

Full Papers
doi.org/10.1002/ejoc.202001469
[12] G. Desimoni, G. Faita, K. A. Jørgensen, Chem. Rev. 2006, 106, 3561–3651.
[13] J. Bakos, Á. Orosz, S. Cserépi, I. Tóth, D. Sinou, J. Mol. Catal. A 1997, 116,
85–97.
[14] a) B. M. Trost, D. L. Van Vranken, Chem. Rev. 1996, 96, 395–422; b) D. A.
Evans, M. M. Faul, M. T. Bilodeau, B. A. Anderson, D. M. Barnes, J. Am.
Chem. Soc. 1993, 115, 5328–5329; c) J. P. Page, T. V. RajanBabu, J. Am.
Chem. Soc. 2012, 134, 6556–6559.
Conflict of Interest
1
2
3
4
5
The authors declare no conflict of interest.
Keywords: Asymmetric catalysis
· Carboxylation · Carbon
[15] a) H. M. Lee, I. S. Youn, M. Saleh, J. W. Lee, K. S. Kim, Phys. Chem. Chem.
Phys. 2015, 17, 10925–10933; b) E. Hernández-Marín, A. A. Lemus-
Santana, J. Mex. Chem. Soc. 2015, 59, 36–42; c) M. Prakash, K. Mathivon,
D. M. Benoit, G. Chambaud, M. Hochlaf, Phys. Chem. Chem. Phys. 2014,
16, 12503–12509.
[16] D. Kajiya, K.-i. Saitow, J. Phys. Chem. B 2010, 114, 16832–16837.
[17] P. Li, J. M. Maier, E. C. Vik, C. J. Yehl, B. E. Dial, A. E. Rickher, M. D. Smith,
P. J. Pellechia, K. D. Shimizu, Angew. Chem. Int. Ed. Engl. 2017, 56, 7209–
7212.
[18] The costly separation of enantiomers of the synthesized racemic
mixture of L2 was not attempted, due to the computationally predicted
low enantiomeric excess expected for this ligand in hydrocarboxyla-
tions.
[19] M. J. Frisch, G. W. Trucks, H. B. Schlegel, G. E. Scuseria, M. A. Robb, J. R.
Cheeseman, G. Scalmani, V. Barone, B. Mennucci, G. A. Peterson, H.
Nakatsuji, M. Caricato, X. Li, H. P. Hratchian, A. F. Izmaylov, J. Bloino, G.
Zheng, J. L. Sonnenberg, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J.
Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T. Vreven,
J. A. Jr Montgomery, J. E. Peralta, F. Ogliaro, M. Bearpark, J. J. Heyd, E.
Brothers, K. N. Kudin, V. N. Staroverov, R. Kobayashi, J. Normand, K.
Raghavachari, A. Rendell, J. C. Burant, S. S. Iyengar, J. Tomasi, M. Cossi,
N. Raga, J. M. Millam, M. Klene, J. E. Knox, J. B. Cross, V. Bakken, C.
Adamo, J. Jaramillo, R. Gomperts, R. E. Stratmann, O. Yazyev, A. J. Austin,
R. Cammi, C. Pomelli, J. W. Ochterski, R. L. Martin, K. Morokuma, V. G.
Zakrzewski, G. A. Voth, P. Salvador, J. J. Dannenberg, S. Dapprich, A. D.
Daniels, O. Farkas, J. B. Foresman, J. V. Ortiz, J. Cioslowski, D. J. Fox,
Gaussian 09, rev. D.01; Gaussian, Inc., Wallingford, CT, 2013.
[20] a) A. D. Becke, Phys. Rev. A 1988, 38, 3098–3100; b) C. Lee, W. Yang, R. G.
Parr, Phys. Rev. B 1988, 37, 785–789.
dioxide fixation · Density functional calculations · Rhodium
6
7
8
9
[1] a) M. Cokoja, C. Bruckmeier, B. Rieger, W. A. Herrmann, F. E. Kühn,
Angew. Chem. Int. Ed. Engl. 2011, 50, 8510–8537; b) S. Dabral, T. Schaub,
Adv. Synth. Catal. 2019, 361, 223–246; c) A. Tortajada, F. Julia-Hernandez,
M. Borjesson, T. Moragas, R. Martin, Angew. Chem. Int. Ed. 2018, 57,
15948–15982; d) Y. Yang, J.-W. Lee, Chem. Sci. 2019, 10, 3905–3926;
e) Y.-X. Luan, M. Ye, Tetrahedron Lett. 2018, 59, 853–861; f) J. Vaitla, Y.
Guttormsen, J. K. Mannisto, A. Nova, T. Repo, A. Bayer, K. H. Hopmann,
ACS Catal. 2017, 7, 7231–7244.
[2] a) T. Fujihara, K. Nogi, T. Xu, J. Terao, Y. Tsuji, J. Am. Chem. Soc. 2012,
134, 9106–9109; b) T. León, A. Correa, R. Martin, J. Am. Chem. Soc. 2013,
135, 1221–1224; c) S. Kawashima, K. Aikawa, K. Mikami, Eur. J. Org.
Chem. 2016, 3166–3170; d) C. M. Williams, J. B. Johnson, T. Rovis, J. Am.
Chem. Soc. 2008, 130, 14936–14937; e) M. D. Greenhalgh, S. P. Thomas,
J. Am. Chem. Soc. 2012, 134, 11900–11903; f) P. Shao, S. Wang, C. Chen,
C. Xi, Org. Lett. 2016, 18, 2050–2053; g) K. Murata, N. Numasawa, K.
Shimomaki, J. Takaya, N. Iwasawa, Chem. Commun. 2017, 53, 3098–
3101; h) M. Juhl, S. L. R. Laursen, Y. Huang, D. U. Nielsen, K. Daasbjerg, T.
Skrydstrup, ACS Catal. 2017, 7, 1392–1396; i) K. Ukai, M. Aoki, J. Takaya,
N. Iwasawa, J. Am. Chem. Soc. 2006, 128, 8706–8707; j) H. Mizuno, J.
Takaya, N. Iwasawa, J. Am. Chem. Soc. 2011, 133, 1251–1253.
[3] a) L. Dian, D. S. Müller, I. Marek, Angew. Chem. Int. Ed. 2017, 56, 6783–
6787; Angew. Chem. 2017, 129, 6887–6891; b) Y.-Y. Gui, N. Hu, X.-W.
Chen, L. L. Liao, T. Ju, J.-H. Ye, Z. Zhang, J. Li, D.-G. Yu, J. Am. Chem. Soc.
2017, 139, 17011–17014; c) X.-W. Chen, L. Zhu, Y.-Y. Gui, K. Jing, Y.-X.
Jiang, Z.-Y. Bo, Y. Lan, J. Li, D.-G. Yu, J. Am. Chem. Soc. 2019, 141, 18825–
18835; d) Y. Shi, B.-W. Pan, Y. Zhou, J. Zhou, Y.-L. Liu, F. Zhou, Org.
Biomol. Chem. 2020, 18, 8597–8619; e) C.-K. Ran, X.-W. Chen, Y.-Y. Gui, J.
Liu, L. Song, K. Ren, D.-G. Yu, Sci. China Chem. 2020, 63, 1336–1351.
[4] Y. G. Liu, Q. Chen, C. L. Mou, L. T. Pan, X. Y. Duan, X. K. Chen, H. Z. Chen,
Y. L. Zhao, Y. P. Lu, Z. C. Jin, Y. R. Chi, Nat. Commun. 2019, 10:1675, 1–8.
[5] Lj. Pavlovic, J. Vaitla, A. Bayer, K. H. Hopmann, Organometallics 2018, 37,
941–948.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
[21] S. Grimme, J. Comput. Chem. 2006, 27, 1787–1799.
[22] a) E. Cances, B. Mennucci, J. Tomasi, J. Chem. Phys. 1997, 107, 3032–
3041; b) J. Tomasi, B. Mennucci, E. Cances, J. Mol. Struct. 1999, 464, 211–
226; c) J. Tomasi, B. Mennucci, R. Cammi, Chem. Rev. 2005, 105, 2999–
3094.
[23] R. Krishnan, J. S. Binkley, R. Seeger, J. A. Pople, J. Chem. Phys. 1980, 72,
650–654.
[6] D. García-López, Lj. Pavlovic, K. H. Hopmann, Organometallics 2020, 39,
[24] P. J. Hay, W. R. Wadt, J. Chem. Phys. 1985, 82, 299–310.
[25] A. W. Ehlers, M. Böhme, S. Dapprich, A. Gobbi, A. Höllwarth, V. Jonas,
K. F. Köhler, R. Stegmann, A. Veldkamp, G. Frenking, Chem. Phys. Lett.
1993, 208, 111–114.
[26] a) C. J. Cramer, Essentials of computational chemistry: theories and
models, J. Wiley, West Sussex, England; New York, 2002; b) K. H.
Hopmann, Organometallics 2016, 35, 3795–3807.
1339–1347.
[7] Q. Peng, F. Duarte, R. S. Paton, Chem. Soc. Rev. 2016, 45, 6093–6107.
[8] a) K. H. Hopmann, Int. J. Quantum Chem. 2015, 115, 1232–1249; b) H. J.
Davis, R. J. Phipps, Chem. Sci. 2017, 8, 864–877.
[9] a) H. B. Schlegel, J. Comput. Chem. 1982, 3, 214–218; b) H. B. Schlegel,
Wiley Interdiscip. Rev.: Comput. Mol. Sci. 2011, 1, 790–809; c) J. J. Zheng,
M. J. Frisch, J. Chem. Theory Comput. 2017, 13, 6424–6432; d) C.
Gonzalez, H. B. Schlegel, J. Chem. Phys. 1989, 90, 2154–2161.
[10] a) Y. Guan, V. M. Ingman, B. J. Rooks, S. E. Wheeler, J. Chem. Theory
Comput. 2018, 14, 5249–5261; b) A. R. Rosales, J. Wahlers, E. Limé, R. E.
Meadows, K. W. Leslie, R. Savin, F. Bell, E. Hansen, P. Helquist, R. H.
Munday, O. Wiest, P.-O. Norrby, Nat. Can. 2019, 2, 41–45.
[11] a) F. S. P. Cardoso, K. A. Abboud, A. Aponick, J. Am. Chem. Soc. 2013,
135, 14548–14551; b) P. Braunstein, F. Naud, S. J. Rettig, New J. Chem.
2001, 25, 32–39; c) B. V. Rokade, P. J. Guiry, ACS Catal. 2018, 8, 624–643.
[27] S. T. Schneebeli, M. L. Hall, R. Breslow, R. Friesner, J. Am. Chem. Soc.
2009, 131, 3965–3973.
Manuscript received: November 9, 2020
Revised manuscript received: December 14, 2020
Accepted manuscript online: December 18, 2020
Eur. J. Org. Chem. 2021, 663–670
www.eurjoc.org
670
© 2020 The Authors. European Journal of Organic Chemistry published
by Wiley-VCH GmbH